Re: [Lib2geom-devel] Subpath from path by length?

[mathog <ma...@ca...>]

This may not be the most efficient way to do this, but it works pretty 
well
(sorry about any wrapping that may occur when it is posted):

        // convert the path, gets its complete length, and then make a 
new path with parameter length instead of t
        Geom::Piecewise<Geom::D2<Geom::SBasis> > tmp_pathpw;  // pathv-> 
sbasis
        Geom::Piecewise<Geom::D2<Geom::SBasis> > tmp_pathpw2; // sbasis 
using arc length parameter
        Geom::Piecewise<Geom::D2<Geom::SBasis> > tmp_pathpw3; // new 
(discontinuous) path, composed of dots/dashes
        int n_dash = style->stroke_dash.n_dash;
        int i=0; //dash index
        double tlength;                                       // length 
of tmp_pathpw
        double slength=0.0;                                   // start 
of gragment
        double elength;                                       // end of 
gragment
        for (unsigned int i=0; i < pathv.size(); i++) {
           tmp_pathpw.concat(pathv[i].toPwSb());
        }
        tlength = length(tmp_pathpw,0.1);
        tmp_pathpw2 = arc_length_parametrization(tmp_pathpw);

        std::cout << "length of pathpw:"<< tlength << std::endl;
        // go around the dash array repeatedly until the entire path is 
consumed (but not beyond).
        while(slength < tlength){
           elength = slength + style->stroke_dash.dash[i++];
           if(elength > tlength)elength = tlength;
           Geom::Piecewise<Geom::D2<Geom::SBasis> > 
fragment(portion(tmp_pathpw2, slength, elength));
           slength = elength;
           slength += style->stroke_dash.dash[i++];  // the gap
           if(i>=n_dash)i=0;
           tmp_pathpw3.concat(fragment);
        }
        Geom::PathVector out_pathv = 
Geom::path_from_piecewise(tmp_pathpw3, 0.01);

The only thing it does poorly is the handling of the dash located at 
P.begin(), which is always broken into
two pieces. When viewed in inkscape it looks better, as if it was 
wrapped around that corner.  In terms of efficiency this code may not be 
so good since the portion() method presumably has to repeat position 
calculations from 0 to the previous elength, since it does not save the 
end point of the previous fragment.

Anyway, good enough.  Now on to the boolops, which is going to be 
harder since the underlying 2geom function
is apparently broken...

Thanks,

David Mathog
ma...@ca...
Manager, Sequence Analysis Facility, Biology Division, Caltech